Abstract
In the design process of product art modeling image, this paper uses the fuzzy cognitive theory of output variables to do artificial intelligence operations on the products art design, which achieves a virtual intelligent simulation. The decoupling process of artificial intelligence uses MATLAB to code, and establishes a data connection using MATLAB toolbox and Solidworks, which realizes the visualization operation in the design process. Through the analysis of the system testing and design case, artificial intelligent algorithms can effectively realize cognitive thinking storage, and play a feedback effect in product shape artistic image optimization design, which provides a technical reference for the study on product art design.
Introduction
In traditional industrial design, the "artistic" modeling of a product—such as a CNC machine tool—is often a manual and subjective process. While engineering constraints are easily calculated, aesthetic appeal is significantly harder to quantify. This paper identifies a critical gap in how modern industrial design systems handle the transition from conceptual art to functional 3D models. Specifically, there is a need for a system that can "learn" design preferences and automatically adjust parameters to reach an optimal visual result. The challenge lies in "decoupling" complex design variables so that an AI can manipulate them effectively within a virtual simulation environment.
Methodology
The proposed "Network Collaborative Modeling" approach relies on three main pillars: Fuzzy Cognitive Theory, Intelligent Decoupling Algorithms, and a Joint Simulation environment.
System Architecture
The practical core of this work is the integration of MATLAB and Solidworks. MATLAB serves as the "brain," handling AI coding and neural network operations, while Solidworks provides the "body" through a virtual reality environment. A data interface allows MATLAB to send real-time commands to update the 3D model, creating a direct visualization of the AI's optimization process.
Intelligent Decoupling and Neural Networks
To manage the interplay between different design elements, the authors developed discrete mathematical expressions that allow the system to isolate and adjust specific proportional and integral parameters during simulation. The study utilized a Back-Propagation (BP) neural network trained on 46 samples to recognize characteristic values. While the primary focus is CNC tools, the network was validated using road characteristic values and vehicle types to prove its classification and approximation capabilities.
Key Results and Insights
The system demonstrates a powerful ability to "store" cognitive thinking by using neural networks to maintain a memory of design traces. This creates a "standard reference system" that the AI uses to learn what constitutes a high-quality design over time.
Feedback Regulation
A central feature of the framework is the feedback regulation function, which closes the loop between virtual simulation and practical output. The AI does not merely generate a design; it constantly compares the output against a desired "cognitive image" and self-corrects.
The BP neural network achieved a 100% correct rate in training for the sample data, demonstrating that the algorithm is highly effective at approximating desired output values. The authors successfully demonstrated a "cognitive image artificial intelligence process" where raw processing routes are refined into optimized design points, resulting in sleek, ergonomic product forms.
Limitations and Future Work
While the framework is theoretically sound and successful in simulation, its practical application in complex industrial settings requires further investigation. A notable limitation is the jump in the study between CNC machine tool design and vehicle identification data. Future research should focus on scaling the dataset and specifically encoding artistic preferences for industrial machinery into the neural network to bridge this gap.
Conclusion
By combining fuzzy cognitive theory in MATLAB with virtual reality modeling in Solidworks, this research provides a robust framework for automating aesthetic design. The system's ability to decouple complex variables and utilize feedback loops offers a promising technical reference for the future of intelligent product art design.
Summary updated